| name | deep-thoughts |
| description | Use when the user wants rigorous thinking before landing on an answer โ framing a fuzzy problem, surfacing hidden assumptions, stress-testing a decision, running a premortem, steelmanning an opposing view, applying five-whys root-cause analysis, reframing a question, doing first-principles decomposition, finding a double-crux, or deliberating an ambiguous open-ended choice. Especially applies to strategy, product direction, research design, architecture decisions, hiring calls, ethical tradeoffs, and any prompt where jumping straight to a solution would skip important questions. Trigger phrases include "help me think through", "should I", "what should we build", "what am I missing", "reframe this", "pre-mortem", "steelman", "assumption audit", "debug my thinking", "five whys", "stress-test this decision", "first principles".
|
| version | 0.1.0 |
| author | jon-chun |
| license | MIT |
| homepage | https://github.com/jon-chun/skills-toolbox |
| keywords | ["reasoning","socratic","design-thinking","premortem","steelman","first-principles","critical-thinking","decision-quality"] |
Deep Thoughts
Overview
A disciplined thinking companion. When a user presents a fuzzy problem, a high-stakes decision, or an ambiguous open-ended question, slow down before answering: surface the assumptions, reframe the question if needed, and deliberate with a concrete Socratic or design-thinking method before producing a recommendation.
Core principle: the quality of an answer is capped by the quality of the question. Spend the first third of effort on the question itself.
When to use
Trigger on any of:
- Strategic / product / research decisions with unclear success criteria
- Problems the user is "stuck" on or keeps circling
- Requests to "think through", "help me decide", "what should we build", "is this a good idea"
- Hiring, ethics, tradeoffs, or reputationally-sensitive choices
- Plans that feel "obviously right" โ overconfidence is a trigger
- Post-mortems about near-misses that almost went badly
When NOT to use
Skip this skill when:
- The user needs a direct factual answer ("what port does SSH use?")
- There is already a clear, shared success criterion and the work is execution
- Speed matters more than deliberation (outage, live-site incident)
- The user has explicitly asked for a quick answer or a vibe check
Core methods โ quick reference
| Method | When to reach for it | Output shape |
|---|
| Socratic ladder | Vague goal, unclear problem | A tree of clarifying โ probing โ evidential โ implication questions |
| Five Whys | Recurring symptom, need root cause | A chain of 5+ "why?" with a named root cause |
| Assumption audit | Plan feels obvious or fragile | Ranked list of assumptions with confidence and consequence-if-wrong |
| Premortem | Before committing resources | "Imagine this failed โ why?" โ ranked failure modes with mitigations |
| Steelman | Disagreement, contested view | The strongest version of the opposing view, built charitably |
| Double-crux | Two parties disagree and are stuck | A single sub-question whose answer would change both minds |
